Structural Evolution under Pressure of BiMnO3 - Inorganic Chemistry

Chem. , 2014, 53 (16), pp 8749–8754. DOI: 10.1021/ic5013878. Publication Date (Web): July 31, 2014. Copyright © 2014 American Chemical Society. *E-...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/IC

Structural Evolution under Pressure of BiMnO3 Gianluca Calestani,*,† Fabio Orlandi,† Francesco Mezzadri,† Lara Righi,†,‡ Marco Merlini,§ and Edmondo Gilioli‡ †

Dipartimento di Chimica, Università degli Studi di Parma, Parco Area delle Scienze 17A, I-43124 Parma, Italy CNR-IMEM, Parco Area delle Scienze 37A, I-43124 Parma, Italy § Dipartimento di Scienze della Terra, Università degli Studi di Milano, Via Botticelli 23, I-20133 Milano, Italy ‡

S Supporting Information *

ABSTRACT: The structural behavior of BiMnO3 under pressure was studied in a quantitative way by single-crystal synchrotron Xray diffraction up to 36 GPa. Two phase transitions were observed at moderate pressures (1 and 6 GPa, respectively), leading the system at first to the P21/c and then to the Pnma symmetry. The breaking of C-centering in the first transition does not affect significantly Jahn−Teller (JT) distortion and orbital order (OO) but determines a significant change in the cooperative tilting of the MnO6 octahedra. The second transition increases the symmetry to orthorhombic, leading to a Pnma structure similar to the O′ structure of LaMnO3, characterized by a > c > b/√2. No symmetry change was observed above 7.1 GPa, but the different compressibility of the lattice parameters (in particular, the b axis) leads at first to a pseudocubic phase (≈30 GPa) and then to different parameter ratios (b/√2 > c > a). Even if the JT distortion is continuously reduced with increasing pressure, it is retained, together with the resulting OO, until the highest measured pressure, pointing out the relevant role of the distortion induced by the Bi3+ lone pair in stabilizing the JT distortion.



INTRODUCTION Magnetoelectric multiferroic materials,1 combining electric and magnetic order, attract attention in solid-state physics, also in view of possible applications in new electronic devices.2−6 Among the various magnetoelectrics reported in recent years, BiMnO3, synthesized under high pressure,7,8 is a rare example of ferromagnetic (FM) material. The FM ordering, observed at Tc = 100 K in this system,9 originates by superexchange10,11 taking place in the presence of a peculiar orbital ordering (OO) resulting from a heavily distorted perovskite structure. Compared to LaMnO3, which has an antiferromagnetic (AFM) ground state also originating from OO and superexchange,12,13 the FM state in BiMnO3 is at first sight surprising and points out the relevance of the distortion induced by the Bi3+ 6s2 lone pair in determining the peculiar OO pattern. In recent years, both the structure and the properties of BiMnO3 have been a matter of debate, some doubt being expressed as to whether it is multiferroic at all. It has been generally accepted that the BiMnO3 phase I, metastable at room temperature (RT) and ambient pressure (AP), crystallizes in the noncentrosymmetric space group C2.14,15 However, a centrosymmetric C2/c structure, incompatible with ferroelectricity (FE), was later suggested by first-principles calculations16 and confirmed by experimental studies17,18 performed not only at RT but also below the magnetic transition. More recently, Solovyev and Pchelkina19−21 put forward the idea that FE in BiMnO3 could be improper, being related to the competition between the FM © 2014 American Chemical Society

state (with C2/c symmetry, deriving from conventional nearestneighbor interactions) and a hidden ↑↓↓↑ AFM state, which breaks the inversion symmetry (produced by the peculiar OO via longer-range interactions between remote Mn atoms). The idea that the structure could be rather instable and could be easily modified by small variations of external parameters, suggested by the competition of FM and AFM states, found further support in the complex structural behavior of BiMnO3. A minority polymorph with a higher magnetic TC (107 K) was found to coexist at RT with phase I in the as-prepared material and to disappear on heating,22 where two reversible transitions were observed at 470 and 770 K.14,23 The first one leads to a further monoclinic phase (II), whose structure was determined in C2/c by Belik el al.,17 whereas the latter to a LaMnO3-related orthorhombic phase (III). The structure of BiMnO3 is known also to depend on its oxygen stoichiometry.24,25 By increasing the Mn4+ content (produced by cation deficiency) indeed, the symmetry evolves again to orthorhombic Pnma through two different monoclinic modifications (C2/c and P21/c, respectively), each one characterized by peculiar magnetic properties and decreasing TC. A similar structural trend was observed by increasing the pressure26,27 up to 10 GPa: a new monoclinic phase was found to be stable above 1 GPa and to convert to the orthorhombic Pnma one in the 6−8 GPa range. The symmetry Received: June 18, 2014 Published: July 31, 2014 8749

dx.doi.org/10.1021/ic5013878 | Inorg. Chem. 2014, 53, 8749−8754

Inorganic Chemistry

Article

Table 1. Crystal Data for the BiMnO3 Structure at Different Pressures P (GPa)

phase

system

s.g.

a (Å)

b (Å)

c (Å)

β (deg)

V (Å3)

Dx (g/cm3)

0.35 1.2 2.9 4.5 7.1 9.4 11.6 15.2 18.2 25.7 30.1 35.3

I P P P O′ O′ O′ O′ O′ O′ O′ O′

monocl. monocl. monocl. monocl. orthorh. orthorh. orthorh. orthorh. orthorh. orthorh. orthorh. orthorh.

C2/c P21/c P21/c P21/c Pnma Pnma Pnma Pnma Pnma Pnma Pnma Pnma

9.513(3) 9.6230(13) 9.5921(5) 9.5619(6) 5.5282(3) 5.4987(2) 5.4708(6) 5.4294(6) 5.3910(5) 5.3104(7) 5.2708(4) 5.2235(4)

5.5941(5) 5.46038(19) 5.43141(8) 5.40751(9) 7.5899(9) 7.5692(7) 7.5494(15) 7.5233(14) 7.4997(14) 7.4649(18) 7.4467(10) 7.4274(12)

9.836(11) 9.824(4) 9.7664(19) 9.725(2) 5.4641(11) 5.4400(8) 5.4221(19) 5.3913(18) 5.3650(18) 5.306(2) 5.2725(12) 5.241(12)

110.37(8) 110.96(3) 110.872(14) 110.795(14) 90 90 90 90 90 90 90 90

490.7(2) 482.05(8) 475.42(7) 470.07(8) 229.26(8) 226.42(7) 223.94(9) 220.22(10) 216.91(9) 210.32(11) 206.94(8) 203.72(7)

8.44 8.59 8.71 8.81 9.03 9.15 9.25 9.40 9.58 9.85 10.01 10.19

of the intermediate monoclinic phase was determined as P21/c by a synchrotron X-ray powder diffraction experiment,26 whereas, in a further neutron diffraction study, the structure was refined in C2/c.27 It is interesting to note that the highpressure orthorhombic phase was reported to present an OO similar to that of LaMnO3, which is regarded as the archetypal cooperative Jahn−Teller (JT) and orbitally ordered system. Generally speaking, the JT distortion is expected to be reduced under pressure (potentially until a total suppression), but the details of this evolution as well as the occurrence of an eventual total suppression have been debated.28−30 In fact, it was first claimed, by synchrotron X-ray diffraction, that the JT was completely suppressed in LaMnO3 above 18 GPa,28 before further measurements by Raman and X-ray absorption spectroscopy suggested a persistence of the JT distortion, coexisting with a new emergent undistorted phase, up to 34 GPa well above the insulator-to-metal transition.29,30 The discrepancy with X-ray results suggests that the phase segregation revealed by these techniques occurs on a spatial or time scale that cannot be accessed by X-ray diffraction. Since Bi3+ has an ionic radius very similar to that of La3+ but a stereochemistry that is strongly influenced by its 6s2 lone-pair character, the comparison of the structural behavior of LaMnO3 and BiMnO3 under high pressure is considered quite important in order to understand the possible role of the lone-pairinduced distortion in stabilizing JT distortion and OO. Very recently, two different studies appeared in the literature,31,32 dealing with structural characterization of BiMnO3 at pressures above 10 GPa. In the first one,31 beside a P−T phase diagram obtained by means of energy-dispersive powder X-ray diffraction in the ranges of 0−4 GPa and 300−900 K, the structural behavior at high pressures and ambient temperature of BiMnO3 was studied by means of angle-dispersive powder XRD and Raman spectroscopy up to 50 GPa. At P > 20 GPa, a structural phase transition from Pnma to a new orthorhombic Imma phase was reported. Following the authors, the transition would start at P ∼ 20 GPa, evolving over a pressure range of several GPa, leading to the suppression of the long-range d(3y2−r2) eg OO, being the static cooperative JT distortion forbidden in Imma. Completely different results were obtained by synchrotron XRD and Raman spectroscopy up to 60 GPa.32 The Pnma structure was found to remain remarkably stable between 7 and 37 GPa, and no transition to Imma was observed. On the contrary, the transition to a strongly elongated monoclinic phase was found to occur between 37 and 39 GPa, followed by the transition to a triclinic metallic phase above 53 GPa. On the basis of the Rietveld refinement of

the powder XRD data collected at 50 GPa, the authors claim that the high-pressure monoclinic structure is noncentrosymmetric and allows ferroelectricity, which contradicts the traditional expectation that ferroelectricity vanishes under pressure. However, given the history of the determination of the structure of phase I of BiMnO3, any assignment of noncentrosymmetry, on the basis of powder XRD, should be prudent, especially given the quality of the data available at such high pressures. In the same work, the structural study under pressure was also attempted by using “single” crystals, but the observed twinning made a full structural refinement impossible in the whole pressure range. Concurrently, we performed a similar synchrotron XRD experiment. Unfortunately, our study was limited in pressure to 36 GPa so that a more limited pressure range was explored. However, the data were collected from single crystals that demonstrated to be suitable for similar experiments, offering us the possibility to accurately reinvestigate, in a quantitative way, the structural behavior of BiMnO3 under pressure, in order to clarify aspects that remained controversial after previous investigations.



EXPERIMENTAL SECTION

Diffraction experiments were performed at the ID9A beamline of the ESRF (Grenoble)33 by monochromatic X-ray radiation (λ = 0.41456 Å). Wavelength and geometrical instrumental parameters (sample-todetector distance, beam center, detector tilt) were refined against standard silicon powder and a pure quartz single crystal. BiMnO3 single crystals were extracted from a ceramic sample and prepared by high-pressure−high-temperature synthesis (4 GPa, 1073 K) from a mixture of the binary oxides containing a slight Bi2O3 excess. The crystal quality of small fragments, with dimensions ranging around a few microns, were preliminarily tested on the beamline. A small crystal with dimensions of 5 × 5 × 4 μm3, practically free from the usual twinning arising in BiMnO3 by the phase transitions occurring at the end of the preparation process by the final temperature decrease and pressure release, was selected and mounted in a diamond anvil cell (DAC), used for pressure generation with helium as a pressure medium to provide nearly hydrostatic conditions. Diffraction data suitable for structure solution and refinement were collected at 293 K with a Mar555 flat panel detector in the range of 2.5 < θ < 19° by 0.5° ω step in the angular interval of ±30° for 12 pressure steps ranging from 0.35 to 35.3 GPa. The intensity integration was performed with the Oxford Diffraction software CrysalisRed-171.32.29 (Oxford Diffraction Ltd., Abingdon, England), using the profile fitting algorithm and no model refinement. Empirical rescaling and absorption correction were applied. The structures were solved by direct methods with SIR200434 and refined with anisotropic atomic displacements parameters (a.d.p.) for Bi and Mn atoms by using JANA200635 for all the investigated pressures. 8750

dx.doi.org/10.1021/ic5013878 | Inorg. Chem. 2014, 53, 8749−8754

Inorganic Chemistry



Article

pressure in similar conditions (above 10 GPa) also in BiFeO3,40 but in this case, the cell parameters reveal an almost linear pressure-dependent evolution. The analysis of the Mn−O bond distances at the 1.0 GPa transition indicates that the breaking of C-centering does not affect significantly the JT distortion. The OO scheme of phase P is similar to that of phase I and is shown schematically in Figure 2. On the contrary, the phase transition

RESULTS Crystal data for all the investigated pressures are reported in Table 1. Refinement parameters, atomic coordinates, and Mn− O bond distances are reported as Supporting Information in Tables SI and SII, respectively, for four selected examples that are representative of the observed behavior. As expected by previous investigations,26,27,31,32 two phase transitions are clearly observed around 1.0 and 6.0 GPa, respectively. However, differently from what was reported for powder samples, the transitions are quite sharp and no coexistence of different phases was observed in our single-crystal experiments. This suggests that the wide phase coexistence previously observed in powder experiments is not intrinsic but likely originates from stress between grains or local inhomogeneities, which, in turn, underlines the importance of the single-crystal experiments we performed. The first transition, keeping the monoclinic symmetry, breaks the C-centering typical of phase I. The P21/c space group was assigned to the new phase (hereinafter indicated as P) on the basis of systematic absences. This result agrees with previously published synchrotron data26,31 and indicates that the use of C2/c symmetry to refine the structure by neutron data27 was probably dictated by insufficient resolution or poor data quality. The second transition (P = 6.0 GPa) increases the symmetry to orthorhombic, leading to a Pnma structure similar to the O′ structure of LaMnO3,36−38 characterized by a > c > b/√2. According to Table 1, no symmetry change was observed above 7.1 GPa, but the different compressibility of the lattice parameters (in particular, of the b axis) leads at first to a pseudocubic phase (≈30 GPa) and then to different parameter ratios (b/√2 > c > a). This is shown in Figure 1, where the

Figure 2. Orbital ordering in BiMnO3 structures: (a, b) monoclinic phases (I and P); (c, d) orthorhombic Pnma phase. Only the occupied eg orbitals of Mn3+ are shown; ap, bp, and cp refer to the fundamental perovskite axes.

determines a significant change in the cooperative tilting of the MnO6 octahedra. As shown in Figure 3, the in-plane tilt in adjacent perovskite layers, which propagates with an opposite phase in I, becomes in-phase in P so that the final transition to O′ simply involves the equivalence of the adjacent layers by the increased symmetry. As shown in Figure 4, the structural rearrangement involved into the I−P transition results in a significant change of the Mn−O−Mn angles, which become dispersed over a wider range. The Mn−O−Mn angles tend to decrease by increasing pressure, before the sudden collapse in the 148−150° range at the P−O′ transition. The structure of O′-BiMnO3 is similar to that of LaMnO3, even if an important shift of the O(2) atom position is produced in BiMnO3 by the necessity to accommodate the Bi3+ 6s2 lone pair. The strong JT distortion observed in the monoclinic phases survives in O′, but the pattern of the occupied eg orbitals changes drastically and, in analogy with LaMnO3, develops parallel to the ac plane, as shown in Figure 2. Being the two compounds characterized by similar OO and space symmetry, the same magnetic behavior of LaMnO3, i.e., A-type AFM, is, in principle, expected for orthorhombic BiMnO3. The pressure evolution of the O′ structure is resumed in Figure 5. The linear decrease of the Mn−O distances, observed with increasing pressure, suggests a progressive reduction of the cooperative JT effect, a process that could be completed at about 45 GPa, if the observed trends would be maintained beyond the maximum applied pressure: in this case, at this pressure, all the bond lengths are expected to become nearly equivalent by extrapolation of the observed trends. It is interesting to note that this pressure is much higher than the one required for the observation of OO quenching in LaMnO3 by XRD (18 GPa),

Figure 1. Pressure dependence of (right axis) BiMnO3 cell volume, normalized to the fundamental perovskite lattice to compare different phases (the values of the HP monoclinic phase are taken from ref 32) and (left axis) lattice parameters of the orthorhombic cell occurring at higher pressure; b-axis data are normalized by a factor of (1/√2), corresponding to a pseudocubic lattice representation.

pressure dependence of the lattice parameters of the orthorhombic structure and the behavior of the cell volume (normalized to the fundamental perovskite lattice in order to compare different phases) is reported. It was found that the V(P) data of the orthorhombic structure are well described by the Birch−Murnaghan equation39 and the derived bulk modulus K0 and its pressure derivative K′0 at zero pressure are reported in Figure 1. It is interesting to note that an orthorhombic Pnma modification has been observed under 8751

dx.doi.org/10.1021/ic5013878 | Inorg. Chem. 2014, 53, 8749−8754

Inorganic Chemistry

Article

Figure 3. Pressure-dependent evolution of the BiMnO3 structure (from left to right: phase I, P, and O′), projected along the same fundamental perovskite axis (determined by the analysis of the orientation matrices). Arrows show schematically the in-plane tilt of MnO6 octahedra in layers alternating along the view axis. Different colors refer to symmetry-independent octahedra.

Figure 4. Pressure dependence of the Mn−O−Mn angles. In red, black, and blue are reported the data corresponding to the C2/c, P21/c, and Pnma phases, respectively.

pointing out the relevant role of the Bi3+ lone pair in stabilizing the JT distortion of trivalent manganese. Quite interesting is also the behavior of the Mn−O−Mn angles, which increase under pressure in the orthorhombic phase, at least until the pseudocubic minimization of the lattice parameter distortion is reached. The increase is more evident for the angle involving the apical O(1) atom, and this is probably driven by the decrease of the apical Mn−O(1) distance. The two Mn−O− Mn angles are frozen above 25 GPa, and the structural stress cumulated in this region could be at the origin of the further transition observed above 37 GPa. Our structure analysis shows that the distortion of the Bi environment, originated by the steric requirements of the lone-pair accommodation, persists in the whole range analyzed in spite of the increase of the number of shorter Bi−O interactions (